The present invention relates to image processing method and image processing programs using computer graphics technology.
The method rendered in JP-A-2000-251094 has been already known as an image processing method to render the outline of a model emphatically. In this method, a first-model and a second-model, slightly bigger than the first one, are prepared to render one model. A normal vector of a polygon constituting the second-model is made turn over to the normal vector of the polygon corresponded to the first-model. The first-model is rendered so that it is contained by the second-model. At the moment of the rendering, a hidden surface (the hidden part in view of the view point) of each model is specified according to a inner product of the normal vector and an eyes vector of each model, the specified hidden surface is excluded from the object to be rendered. Therefore, about the second model, only the outside part of the first-model is rendered. As a result, the second-model is visible as the outline of the first model.
The above mentioned method is that each model is divided into the parts to be rendered or not to be rendered according to the normal vector of each polygon, after that, only the part to be rendered is rendered on a drawing memory. Therefore, the processing amount when models are rendered on the drawing memory (rendering) decreases, but the processing amount before the rendering increases. On a conventional image processing unit carried by a game machine and the like, compared with the drawing processing, the unit has the high ability of a vector operation processing and the like as the preprocessing, therefore, as above mentioned, it is effective for the improvement of processing speed that the normal vector operation for every model and the like are executed in order to reduce the number of polygons to be rendered in the memory.
However, by the improvement of the computer-processing ability of late years the performance of the rendering engine carried by game machines has been extremely improved. So, it is sometimes more advantageous for the improvement of processing speed to exclude such as the preprocessing as much as possible than to execute the special preprocessing to reduce the number of polygons to be rendered.
Furthermore, in the method of the bulletin, as the processing to make the normal vector of the second-model turn over based on the polygon-data of the first-model is essential, the shape of the second-model is restricted to the shape with some geometric relation (for typical example, a similar figure) to the first-model. So, the flexibility of representation by the second-model is small.
In the image processing using computers, a z-buffer algorithm has been already known as a method to render a plural number of models arranged in the virtual 3-dementional space while maintaining their order relations correctly. In this method, the special buffer-memory area (called z-buffer) having the drawing area which is similar to a frame-buffer is reserved in the drawing memory and in the z-buffer a depth information in view of the view point of a model rendered in the frame-buffer is recorded by every pixel. When the new model is rendered in the frame-buffer, the depth information of each pixel of the model for rendering is compared with the depth information recorded in the z-buffer. Only the part of the model for rendering where the depth information is smaller (i.e. closer in view of the view point) than the one in z-buffer is newly rendered in the frame-buffer. As this z-buffer algorithm doesn't require the vector operation performed in the hidden-surface removal method using the normal vector, the judgment of the order-relation can be executed at high speed.